Extraction Algorithm of Similar Parts from Multiple Time-Series Data of Cerebral Blood Flow

• Published in: Brain and Health Informatics pp. 138 - 146
• Main Authors: Hiroyasu, Tomoyuki, Fukushma, Arika, Yamamoto, Utako
• Format: Book Chapter
• Language: English
• Published: Cham Springer International Publishing
• Series: Lecture Notes in Computer Science
• Subjects: functional Near-infrared Spectroscopy; Similar Parts; Time-series data
• ISBN: 9783319027524; 3319027522
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/978-3-319-02753-1_14

We propose an algorithm to extract similar parts from two different time-series data sets of cerebral blood flow. The proposed algorithm is capable of extracting not only parts that are exactly the same but also similar parts having a few differences since time-series data of cerebral blood flow is reported to be affected by various factors, and real data may therefore differ from a model system. To confirm the effectiveness of the proposed algorithm, we evaluated two sets of time-series data of cerebral blood flow: one artificial and one of actual data, and evaluated the results by visual confirmation as well as correlation coefficient analysis. This demonstrated that the proposed algorithm was able to extract similar parts from time-series data of cerebral blood flow. We also found that a Low-pass filter was needed to process time-series data of cerebral blood flow, when the data contained high-frequency noise.